Deetech vs FRISS: Modern Insurance Fraud Prevention

FRISS is a well-established fraud detection platform serving the property and casualty insurance market. With real-time risk scoring, network analysis, and automated fraud detection deployed across carriers globally, FRISS has earned its place in the insurance technology stack.

Deetech detects AI-generated and manipulated media in insurance claims — a capability that FRISS was not designed to provide.

Like the comparison with Shift Technology, this is not a competitive analysis. Deetech and FRISS address different fraud vectors. But understanding exactly where each platform’s capabilities begin and end is critical for carriers building a comprehensive fraud defense.

FRISS: Real-Time Fraud Risk Scoring

FRISS has built a strong reputation in P&C insurance fraud detection. Their platform operates across the policy lifecycle — underwriting, claims, and SIU — providing consistent fraud risk assessment at each stage.

Core capabilities:

Real-time risk scoring. Every claim receives an automated fraud risk score at the point of first notice of loss (FNOL). Scores are based on hundreds of data points analyzed against known fraud indicators.
Network analysis. FRISS maps relationships between claimants, witnesses, service providers, legal representatives, and claims handlers. Unusual network patterns — the same repairer appearing across unrelated claims, witness overlap between claimants — surface potential fraud rings.
Underwriting fraud detection. Beyond claims, FRISS assesses applications for misrepresentation, non-disclosure, and identity fraud at the point of underwriting.
Automated SIU triage. Risk scores automatically route suspicious claims to special investigations units, reducing manual review of low-risk claims.
Text mining. Analysis of unstructured text in claims notes, adjuster comments, and correspondence for indicators of fraud.
Claims handler alerts. Real-time notifications to adjusters when a claim exhibits characteristics consistent with fraud patterns.

FRISS serves an important function: ensuring that traditional fraud signals in claims data are detected automatically and consistently. Their platform reduces reliance on individual adjuster instinct and institutional knowledge by codifying fraud indicators into automated scoring models.

FRISS analyses structured claims data, policyholder behavior, network relationships, and text. It does not analyze the media submitted with claims.

This creates a specific blind spot:

Photos: A claimant submits AI-generated photos showing property damage, vehicle damage, or personal injury. FRISS scores the claim based on data patterns — policy validity, claim history, network connections, description consistency. The AI-generated photos are not examined because FRISS does not perform computer vision analysis.

Video: Dashboard camera footage, security camera recordings, or video documentation submitted as evidence. FRISS cannot analyze video content for manipulation, deepfake face swaps, or temporal inconsistencies.

Audio: Voice recordings, phone claim submissions, or recorded statements. Voice cloning technology can produce convincing impersonations of policyholders. FRISS does not perform voice authentication or audio forensic analysis.

Documents: AI-generated repair quotes, medical reports, invoices, and statutory declarations. FRISS may flag inconsistencies in amounts or providers through pattern analysis, but cannot determine whether the document itself was produced by generative AI.

The compounding risk:

This blind spot exists in the context of rapidly advancing generative AI. The tools to produce convincing fake claims evidence are:

Freely available — open-source image generators, consumer voice cloning services, AI document generators
Trivially easy to use — no technical expertise required
Continuously improving — each generation produces more convincing outputs
Specifically targeted — tutorials and communities focused on fraud applications exist on the open internet

Sumsub’s 2024 Identity Fraud Report documented a 245% increase in deepfake-related fraud globally. The Coalition Against Insurance Fraud estimates that fraud already costs US insurers over US$80 billion annually — and that figure predates the current wave of AI-enabled fraud techniques.

The state of deepfake fraud in insurance is clear: generative AI is making fraudulent claims evidence cheaper, faster, and more convincing to produce. A fraud detection stack that cannot examine media is increasingly incomplete.

Where Deetech Fits

Deetech provides the media authenticity verification layer that FRISS lacks. While FRISS analyses claims data and claimant behavior, Deetech analyses the actual evidence submitted with claims.

Deetech’s three-layer approach:

Automated screening. Every media item — photo, video, audio file, document — attached to a claim is scanned automatically at submission. This layer operates in seconds, checking for known AI generation signatures, metadata inconsistencies, and obvious manipulation markers.
Enhanced analysis. Items flagged by automated screening receive multi-model forensic analysis. Multiple detection techniques are applied: frequency domain analysis, GAN fingerprint detection, diffusion model signatures, environmental consistency checks, and compression artifact analysis. This runs in under a minute per item, fully automated.
Forensic investigation. High-risk items receive detailed forensic examination producing court-ready reports with chain of custody documentation, methodology disclosure, and statistical confidence intervals.

What Deetech detects:

AI-generated images (Stable Diffusion, Midjourney, DALL-E, Flux, and other generators)
Manipulated photos (splicing, cloning, inpainting, content-aware edits)
Deepfake video (face swaps, temporal manipulation, synthetic generation)
Voice cloning and audio manipulation
AI-generated documents (synthetic invoices, fabricated certificates)
Recycled imagery (reverse matching against known fraud databases)

FRISS + Deetech: The Combined Defense

The optimal architecture for modern insurance fraud detection combines both capabilities:

FRISS provides:

Claims data pattern analysis
Claimant behavior scoring
Network analysis and fraud ring detection
Underwriting risk assessment
Text analysis of claims notes

Deetech provides:

Media authenticity verification
AI-generated content detection
Document authenticity analysis
Voice authentication
Catastrophe event media correlation

Together, they cover the full threat surface:

Fraud Type	FRISS Detection	Deetech Detection
Staged accidents (data patterns)	✅	—
Fraud rings (network analysis)	✅	—
Serial claimants	✅	—
Inflated claims (data inconsistency)	✅	—
AI-generated damage photos	—	✅
Manipulated video evidence	—	✅
Voice cloning on phone claims	—	✅
AI-generated documents	—	✅
Recycled imagery from other claims	—	✅
Inflated claims with fake photos showing greater damage	Partial (amount patterns)	✅ (media analysis)

The last row is particularly important. Claim inflation — where a genuine incident is exaggerated with fabricated or altered evidence — is one of the most common and costly fraud types. FRISS may flag unusual claim amounts, but the “evidence” supporting the inflated amount appears genuine to data-based analysis. Deetech examines the evidence itself.

Integration Architecture

For carriers running FRISS, adding Deetech is architecturally straightforward:

Parallel processing:

Claim submitted → FRISS scores the data → Deetech analyses the media → Combined risk view
Both systems process simultaneously, adding no sequential delay to claims handling

Claims system integration:

Both FRISS and Deetech integrate with major claims platforms (Guidewire, Duck Creek, Sapiens)
Risk scores and media authenticity results appear in the same adjuster interface
No need to choose one platform’s workflow over the other

SIU workflow:

FRISS flags claims based on data patterns → SIU investigation
Deetech flags claims based on media analysis → SIU investigation
Combined signals — high FRISS score AND media authenticity concerns — indicate highest-priority cases

Independence:

FRISS and Deetech operate independently. Neither requires data from the other to function.
No vendor lock-in or dependency between platforms
Either can be replaced or upgraded without affecting the other

What FRISS Carriers Should Consider

If you’re already running FRISS, you have strong protection against traditional fraud patterns. The question is whether that protection extends to AI-enabled fraud using synthetic media.

Diagnostic questions:

What percentage of your claims include photo or video evidence? For motor, property, and liability claims, the answer is typically 80-95%. All of that media is unexamined by FRISS.
How would you detect an AI-generated photo of vehicle damage? If the answer involves an adjuster’s visual inspection, consider that modern AI-generated images routinely fool trained human observers. A 2023 study published in Psychological Science found that AI-generated faces were rated as more trustworthy than real faces by human evaluators.
What’s your exposure to catastrophe event fraud? After major weather events, the volume of claims creates pressure to process quickly. This is exactly when AI-generated evidence is most likely to succeed — because manual scrutiny decreases as volume increases. FRISS handles the data patterns. Who examines the photos?
Are you seeing claims with unusually high-quality documentation? Paradoxically, AI-generated evidence often looks too good — perfectly lit, well-composed, with no metadata inconsistencies in the data record. FRISS wouldn’t flag these because the data looks clean.
What’s your regulatory exposure? APRA and ASIC are increasingly focused on AI-related risks in financial services. A carrier that cannot demonstrate media verification capability may face questions about the adequacy of their fraud controls.

The Economics

Adding media authenticity detection to an existing FRISS deployment is not a replacement cost — it’s an incremental investment covering a previously unprotected attack surface.

Cost context:

Average cost of an undetected fraudulent motor claim in Australia: A$8,000-15,000
Average cost of an undetected fraudulent property claim: A$15,000-50,000
Number of fraudulent claims using AI-generated evidence: growing rapidly, currently estimated at 3-7% of all fraudulent claims and accelerating
Cost of Deetech per-claim screening: a fraction of a single fraudulent claim payout

The ROI calculation is straightforward: if Deetech catches even a small percentage of AI-enabled fraudulent claims that FRISS cannot detect, the investment pays for itself within the first quarter.

For a detailed financial analysis suitable for executive decision-making, the board-level briefing on generative AI fraud provides a full investment case.

The Future of Insurance Fraud Detection

The distinction between data-pattern fraud detection (FRISS) and media authenticity verification (Deetech) will become increasingly critical as generative AI advances.

Short-term (2026-2027):

AI-generated claims photos become indistinguishable from real photos to human observers
Voice cloning quality reaches the point where phone-based claims are vulnerable
AI document generation produces convincing repair quotes and medical certificates

Medium-term (2027-2029):

AI-generated video evidence becomes a practical fraud tool
Coordinated fraud attacks combine clean data patterns (defeating FRISS-type tools) with synthetic media (defeating visual inspection)
Regulatory requirements for media verification become explicit

Long-term (2029+):

Media authenticity verification becomes a standard component of claims processing, alongside data-pattern analysis
Carriers without media verification face both financial exposure and regulatory non-compliance

The carriers that build their media authenticity capability now — while the technology gap favours defenders — will be better positioned than those that wait until the threat is acute and the regulatory mandate is explicit.

Conclusion

FRISS is an effective fraud detection platform for data-pattern analysis, risk scoring, and network detection. Deetech is an effective media authenticity verification platform for deepfake detection, document analysis, and evidence verification.

They are not alternatives. They are layers in a comprehensive fraud defense stack. FRISS examines the data. Deetech examines the evidence. Both are necessary.

If you’re running FRISS and haven’t addressed media authenticity, the deepfake detection FAQ for insurance companies is a good starting point. For comparison with other tools in the market, see the top deepfake detection tools for insurance in 2026.

To learn how deetech helps insurers detect deepfake fraud with purpose-built AI detection, visit our solutions page or request a demo.